An Evaluation Mechanism Based on HDFS in Unstable Network Environment
Xiaoyu Jin, Xilin Ji, Yang Yang, Zhipeng Gao, Xuesong Qiu · 2018
Reliability of distributed storage is a key issue for storage of big data. This paper presents an evaluation mechanism for the popular distributed file system (HDFS). The evaluation mechanism (RST) is based on HDFS in unstable network environment. The RST can be divided into two parts, one part (NE) is that the NameNode evaluates the storage reliability of DataNode, the other part (CE) is that clients evaluate the communication quality between the client and DataNodes. This paper also proposes strategies of data redundancy and file reading and writing based on RST. The experimental results show that the strategy of data redundancy improves the availability of data and the process of data access improves the success rate of communication in unstable network environment.